t Department of Community Medicine, Wayne State. University, Detroit, Michigan 48202, U.S.A.. Selection for age-specific reproduction has produced replicate ...
Heredity 60 (1988) 367—374
The Genetical Society of Great Britain
Received 16 July 1987
Localizing genes that defer senescence in Drosophila melanogaster Leo S. Luckinbill,* Joseph L. Graves,* Allen H. Reedt and Sulada Koetsawang*
* Department of Biological Sciences and Institute of Gerontology, Wayne State University, Detroit,
Michigan 48202, U.S.A.
t Department of Community Medicine, Wayne State University, Detroit, Michigan 48202, U.S.A.
Selection for age-specific reproduction has produced replicate stocks in which life span exceeds that in short-lived controls by about 30 per cent, in unpaired individuals. Crosses between a selected long-lived (L) stock, short-lived (S) stock and a strain with balancer chromosomes were used to create all possible combinations of their chromosomes. The longest and shortest-lived genotypes are found to be (LSL) and (SLS), with other combinations distributed between them approximately according to their first and third chromosomes. Longevity appears to be under polygenic control with contributing elements on all chromosomes. The third chromosome is by far the most influential, accounting for 66 to 72 per cent of the observed variation in females. The first chromosome is less effective. Epistatic interactions are more important in males than females, but are significant only in measurements of single individuals. Some controlling elements for longevity appear to differ in males and females. Crosses of selected stocks with known P and M-cytotype strains show no effect on either sterility or longevity.
INTRODUCTION
Gerontological
studies have resulted in a vast
informational base describing aging in a variety of
organisms. This has led to the generation of numerous empirically based hypotheses about the genetic and physiological controls on longevity. But the actual fine-scale study of the genetics of aging in Drosophila has awaited the application of selection techniques to establish long- and shortlived stocks from a common ancestral gene pool.
Beginning with Rose and Charlesworth
(1981a) and Rose (1984), followed by Luckinbill
et a!. (1984) and Luckinbill and Clare (1985), studies of selected populations have progressed rapidly, exploring various fundamental aspects of the genetics of longevity using the lines developed. This has included the discovery of additivity in F1
populations from crosses to short-lived control stocks as well as a modifying gene-environment interaction (Clare and Luckinbill, 1985; Luckinbill and Clare, 1986) that may have confounded previous studies, and changes in life history characters
Recently, one study has invoked the methods of quantitative genetics in attempting to estimate the number of genetic elements that extend life in selected stocks. Luckinbill et al. (1987) used Lande's (1981) extension of Wright's (1968) reformulation estimating the numbers of effective
factors that differentiate two counter-selected parental populations. Measurement of F1, F2 and backcrosses were made from crosses of long- with short-lived parental lines. Replicate successive sets of crosses yielded the surprising estimate of a single
gene as determining the increased longevity. The estimates of a single factor conflicted, however,
with one significant prediction of that unique
finding; F2 populations from parental lines differentiated by a single gene should exhibit a phenotypic ratio of 1: 2: 1. Yet no such classes
were evident. Thus, though replicate sets of crosses gave strong indication of a single effective factor, at the same time there was some suggestion that a greater number of factors must be involved.
One of the limitations which burden such studies is the extensive assumptions on which the
and physiological parameters correlated with
methods used rely. Of these, perhaps the most
life span (Rose and Charlesworth, 1981b;; Rose et a!., 1984; Service et a!., 1985; Service, 1987;
(1987), is that genes determining the difference
Service and Rose, 1985).
tenuous of all, in the estimates by Luckinbill et al. observed between parental lines must have effects
368
L. S. LUCKINBILL, J. L. GRAVES. A. H. REED AND S. KOETSAWANG
of equal magnitude and violation of that assumption results in the estimate of the number of factors involved being low.
In this study we use an alternative approach to this problem that both reflects on estimates of gene number and at the same time localizes the elements of interest to specific chromosomes. Here
we use nonrecombinant crosses with a balancer stock to create chromosomal substitution lines having all possible combinations of chromosome pairs
from long-lived (L) and short-lived (S) control stocks.
paired populations. Flies were maintained, during measurement, as in Luckinbill and Clare (1985). Measurements of 960 longevities were analyzed by ANOVA. Variance in longevity of populations showed no particular relationship to means and was heteroscedastic, though not markedly so. Variances were resistant to equalization by conven-
tional transformation methods, but analysis of variance is well known to be little influenced by such inequality when cell numbers are equal, as in this design (Scheffe, 1959; Miller, 1986). There-
fore, to show the effects and interactions of
The stock L2 CyO/ In (2LR ) bw VI ds33 K dpbbw VI;
chromosomes in the various genotypes, a model-i (fixed effects) ANOVA was performed on untransformed longevity values using SPSSX (Statistical Package for the Social Sciences) (Nie and Hull,
In(3LR)DcxF, D3/Sb (Bowling Green Stock Center), was used to provide balancer chromosomes in a series of crosses and backcrosses. Stable pen-
Crosses described here mate recently isolated stocks to older marker-bearing stocks that have
METHODS
1979).
centric inversions incorporate the dominant
been in laboratory culture for many years. Experi-
markers Cy/bw"1 and D3/Sb on chromosomes II and II respectively. bw" is commonly called Plum or Pm. Crosses of the balancer stock with either the long-lived (LLL) or short-lived (SSS) stocks replaced the first chromosome with paired (L) or (S) homologues to create the stocks:
ments were conducted, therefore, to determine whther introduced dysgenic effects may alter longevity in crossed populations. Since the long- and short-lived stocks used here are descended from a population collected from a Michigan orchard in 1979, they are most probably of the P-cytotype.
SCyD3 SCyD3
Crosses between these lines in isolating chromosomes could, therefore, induce considerable levels of mutagenesis and/or male recombination.
and LCyD3 LCyD3 Y Pm Sb S Pin Sb Y Pm Sb L Pin Sb These base stocks were then again crossed to (LLL)
or (SSS) lines to replace chromosomes II and III with those from selected lines. Appendix I shows these. Thus, eight possible substitution combinations were created in which chromosomes I, II and III consisted of paired homologues of either (L) or (S). Lines heterozygous for all chromosomes were constructed by crossing the (SLS) and (LSL) lines. Parental lines (LLL) and (SSS) were reconstructed by backcrossing F1 progeny having balancer chromosomes to parental lines. Unreconstruc-
ted parental lines were also measured. We
Balancer stocks are likely to be M-cytotype.
To determine the extent to which this has occur-
red, selected long- and short-lived stocks were reciprocally crossed to both Harwich and Canton-
S, which are respectively known as P- and Mcytotype lines. Longevity was measured in populations of 30 unpaired males and females from each cross and fecundity was examined in 100 or more females from each cross, testing for gonodal (GD) sterility. RESULTS
attempted to avoid homozygosity as much as poss-
ible by crossing large numbers of individuals. Crosses consisted of 50 pairs of individuals raised
in vials. For measurement and maintenance of cultures, all lines were reared at high larval density
in populations of 50 pair/bottle and were
measured for comparison on the third generation after the completion of crosses. Estimates of longevity here include both adult life span and development. The latter varied by just three days among all the lines described. Longevity was first measured in 30 unpaired males or females in each of eight populations and later in
Table 1 shows the untransformed mean and stan-
dard deviation (s.d.) for both paired and unpaired
chromosomal substitution lines. Parental lines shown there and in figs. 1 and 2 are unreconstructed stocks. In females, two reciprocal substitution lines, (SLS) and (LSL), are respectively the shortest and longest lived lines, with the other combinations of the first and third chromosome distributed about evenly between the extremes and parental lines. In fig. 1 the (SLS) combination has a shorter life span than even the short-lived parental line (SSS).
SENESCENCE GENES IN DROSOPHILA
369
MALES
Table 1 (a) Mean and standard deviation of longevity (in days) for single males and females of chromosome substitu-
tion lines. The genotype of lines are shown for major chromosomes I, II and III. (b) Means for genotypes of paired individuals Cl)
Strain
(a)
Average female Genotype longevity s.d.
2
SSS SSL
3
SLS
4
LSS LLS LSL SLL LLL
1
5
6 7 8
(b)
I
SSS
2 3 4 5 6 7 8
SSL SLS LSS LLS LSL SLL LLL
62-40 70-77 54•97 65-67 62-77 74-50 70-07 78-90
47-13 60-10 48-70 46-23 55-03 65-63 56-00 78-33
Average
male
longevity s.d.
15-24 15-83 10-41 12-29
63-57 7353 65-90 63-87 73-23
1266 1455 1670
8-93
0 > > 0 ci)
10-46 14-49
488
7893
11-42
11-53
66-63
990
15-26
8429
1294
10-49
56-33 65-17 68-47
13-64
1823 1356 13-49 14-46 17-65 16-15 16-03
66-40 68-70 74-80 66-20 79-96
1553 16-59 11-38
1515 15-18 19-61 11-19
LONGEVITY
Figure 2 Survival in single males of parental stocks and chromosome substitution isolates is shown.
particular, occupy opposite positions to that expected.
Fig. 3 compares the mean and 95 per cent
confidence intervals for single male and female recombinant lines of table 1. The same trends found in complete survival curves are clear in this comparison; (SLS) and (LSL) are at or near the
extremes with other recombinants distributed Fig. 2 shows a generally similar pattern with the (SLS) and (LSL) as extremes in males and
loosely between according to whether the first and third chromosomes are (L) or (S).
other combinations distributed between them. One line (LSS) is virtually identical to both the (SSS) and (SLS) lines, but has a slightly shorter average life span than the latter. In males the intermediate
reciprocal crosses of the (LSL) and (SLS) substitution lines, show a strong effect in females of one
isolates are more variable in the position that various combinations exhibit, but a strong effect of the third chromosome is also evident. Among the intermediate genotypes, (SLL) and (LSS), in FEMALES
In table 2, heterpzygous lines created from cross. Other means for F1 populations are found close to midparent values, as have been most of the previous crosses of parental lines. Reclaimation of parental values in unreconstructed lines by sub-
stitution lines is generally excellent. Males and females of the short-lived stock are two to three days lower than values in unreconstructed lines MALES
FEMALES
GENOTYPE
U)
GENOTYPE
SSS —
0 > > U:: 0 Cl)
SLS
LSS
SSS
SLS
LLS
SLL
LSS
SSL
LLS —.---
SSL LSL
-LLL
LLL LONGEVITY
60 70 80 90 LONGEVITY
Figure 1 Survival of single females of parental stocks and isogenic chromosomal substitution lines is shown (in days). ) indicate the short-lived parental stock Dotted lines ( and dashed lines (- - -) show long-lived parent.
Figure
SLL LSL
50
60
70 80
LONGEVITY
3 The mean and 95 per cent confidence interval is
compared for single males and females of all lines.
L. S. LUCKINBILL, J. L. GRAVES, A. H. REED AND S. KOETSAWANG
370
Table 2 Mean and standard deviation of longevity (days) in heterozygotes from crosses of substitution lines SLS with LSL Females Cross
Mean
SSS Parental
6O27
7153
SLS Males x LSL Females LSL Males x SLS Females LLL Parental
Males
Mean
s.d.
591
5960
1223
1153
7497
806
s.d.
(table 1(a)) and long-lived stocks are within about 35 days of their unreconstructed means. Table 3 shows the comparative contribution of all three chromosomes in the ANOVA of paired and unpaired populations. The variation from the error mean square was less than 2 per cent of the total mean square in all analyses and is excluded
from calculations of per cent variation here. In single females, main effects of chromosomes deter-
mine more than 91 per cent of the observable variation in life span, while interactions contribute
55-33
1115
6986
1115
7533
16'03
8O67
17-53
about 9O per cent by contrast. Within the main effects, chromosome III accounts for more than 72 per cent of the observed variation alone with
Table 3 The effects and interactions of chromosomes as determined from a model-I (fixed effects) three-way factorial ANOVA of single females or males from substitution and parental lines Females
Males
Source
of
%
variation
variation
Error
Ch I Ch II Ch III
Ch I xl!
122
1705
135
—166
1211 240 038 350
Ch I XIII Ch lix III xCh III
001
Total
square
2,10042 16662 8,80882
34560
882 73500
013
107
2277
12,31781
%
F
variation
13.9*0
11
138 2831 3-20
58.2*0
4187
15146
591
71•51
281 007 597
ChI xli
Mean Effect
23 0.1 4.9*
01
1151 7.45
556 072
Mean Effect
767 253
920 482 386
square
F
17008 3,496.07
39527 5,17082 1,42107 92042
d.f. 232
20.6**
23 3Ø.4**
1 1 1
84
1
1
—331
68682
5.4* 4.0*
1-30
8882
0.5
2607
12,34937
1
1
239
* 001 < P